Brain Tumor Detection using an Optimized Deep Learning Approach

Brain tumor poses significant challenges in diagnosis and treatment, necessitating accurate and timely detection methods. This study presents a deep learning approach for the detection of brain tumors from medical imaging data. Here, an optimized Convolutional Neural Network (CNN) strategy is implemented for early-stage Brain Tumor Detection. The CNN classifier considers many hyperparameters i.e., number of batches, learning rate, kernel size, weight and bias which may lead to suboptimal result. Here, Differential Evolution (DE) algorithm optimized CNN (DE-CNN), Harmony Search (HS) algorithm optimized CNN (HS-CNN), and Simulated Annealing (SA) algorithm optimized CNN (SA-CNN) are applied to optimize the hyper parameters of CNN. The training and testing accuracy, validation loss, precision and recall values of these three models are compared to reveal the robust model for brain tumor detection. Through extensive experimentation and validation on real-world clinical MRI images, SA-CNN approach demonstrates superior accuracy i.e., 99.63% as compared to other methods.